A method for large, low-contrast acoustic inverse scattering with Born iterations

It is well known in the breast ultrasound inverse scattering problem that the speed of sound contrast is only +/- 6% of the background. However, at ultrasonic wavelengths, the image domain is roughly 200lambda_min cubed. Standard forward solvers such as the finite-difference time-domain method or the method of moments in conjunction with full linear systems to retrieve the objects through various gradient-based iterative techniques are too expensive to be practical. Here we use the Neumann series solution as the forward solver and a multi-objective covariance-based cost function to estimate objects with contrasts in the range of +/- 20% for both density and compressibility. Born iterations are used to update the solution successively. The largest imaging domain tested was 14lambda_min in 2D. Each of these components is well suited to test larger, low-contrast objects.